EVALUATION OF RAIN GAUGE NETWORK IN NARAYANI BASIN, NEPAL

Abstract

Records of precipitation constitute the basic and vital data required for hydrological investigations. These gauges provide point measurements that are interpolated to estimate the precipitation of the catchment. The establishment of a precipitation gauge needs a large amount of investment and human resources, including construction, operation, and maintenance. Even though a denser precipitation gauge network could provide a more precise estimation of regional precipitation, studies have proved that if precipitation stations have been positioned at optimal sites, very fewer numbers of those stations may be enough to model runoff with as good as or better results than when a high-density network is used. Narayani River is a transboundary river basin originating from the Himalayas and flows through the central part of Nepal to converge to the Ganges River in India. Currently, 74 meteorological stations are operating in this basin which is not insufficient as per different international standard and practices. Advancement of optimization models for determining the number of precipitation gauges and their locations is necessary and they can be determined using different concepts. In this thesis, the concept of Entropy and Transinformation Index is used for finding out the optimum number of station network in the basin. First the marginal entropy for all the stations were calculated to find out the station and location with highest uncertainty which represents the first priority station. This station is then coupled with the remaining stations to compute the pair of station with least transinformation which represents the second priority station. This process continues which finally gives out the priority list of the stations. Based on a certain percent of acceptable redundant information and the verification of the results through the rainfall-runoff model HEC-HMS, the optimum number and the location of rain gauge station network is finalized. This work concludes that the combination of 99 stations is an optimum standard station network considering 80% as the acceptable margin for the redundant information. Data of 8 existing stations are found to be above this acceptable margin and their contribution for the hydrological model are found to be insignificant. 33 gridded point locations are suggested for the establishment of possible new stations. The results also almost satisfy the minimum number of stations suggested by the BIS and WMO and also gives good results with the HEC-HMS model output. As the method applied in this basin is scientific yet simple, so can be used for other similar basins also.